Journal oj Personality and Social
1975, Vol. 31, No. 3, 459-466
Psychology
Generosity in Children:
Immediate and Long-Term Effects of Modeling, Preaching,
and Moral Judgment
J. Philippe Rushton
London School of Economics and Political Science
University of London
The differential effectiveness of a model's words and deeds has recently been
found to be problematic. A 3 X 2 X 2 analysis of variance design examined
these effects further and also that of moral judgment on the donating
behavior of 140 children aged 7 to 11 years immediately and 8 weeks later.
Modeling was highly effective both in the immediate and in the follow-up
study and also proved generalizable across changes in the testing situation.
Thus, it was concluded that "internalization" had occurred. Preaching also
proved to be highly effective in the long run although not in the short run.
Some support for a moral judgment relationship with generosity was found.
There has recently been an increasing interest in the determinants of altruistic behavior
in children (Bryan & London, 1970). Much
of this interest has stemmed from a social
learning perspective (Bandura, 1969) involving attempts to test the limits of this paradigm. Within this approach is the recent
series of studies designed to assess the differential effectiveness of words and deeds
upon children's subsequent sharing behavior
(Bryan, 1971; Bryan & Walbeck, 1970a,
1970b; Grusec, 1972; Grusec & Skubiski,
1970).
The prototype experiment used in these
studies involved a bowling game situation
within which a child was able to win valued
tokens, later exchangeable for prizes. At the
same time he was given an opportunity to
donate anonymously some of his winnings to
a charity. The degree to which the child did
This research was carried out in partial fulfillment
of a doctoral dissertation submitted to the University
of London, October 1973. It was supported by a
Social Science Research Council of Great Britain
Postgraduate Award to the author.
The author wishes to acknowledge his appreciation to his supervisor, Professor Hilde T. Himmelweit, for her advice and encouragement in the carrying out of this research and to Ann Holton, Chris
Wilson, and Maryon Hogg who acted as models and
the second experimenter, respectively.
Requests for reprints should be sent to J. P.
Rushton, who is now at the Department of Psychology, York University, 4700 Keele Street, Toronto,
Canada M3J 1P3.
donate constituted the dependent variable of
generosity. Just prior to the child playing the
game, however, he was exposed to a model
playing the same game who either did or
did not donate to the charity and did or did
not preach that one should donate to the
charity.
The results of these studies showed that
while a model's behavior was a potent source
of immediate influence, his preaching was
much more problematic in this respect.
Bryan and Walbek (1970a, 1970b) showed
that while a model's practices influenced the
child's behavior, the same model's preachings
did not. However, both the model's preachings and his behavior affected the child's
judgments of the model's attractiveness.
Grusec and Skubiski (1970) showed that
while a model's behavior was a clear source
of behavior change for both sexes, regardless
of whether the model was nurturant or not,
the model's verbalizations were only effective for females who had been exposed to a
nurturant model. Grusec (1972), however,
did show that a model's verbalizations could
be as effective an influence on the child's
subsequent behavior as a model's behavior,
although not for 7-year-old boys. Clearly,
there is a need to investigate further the
combined and separate effects of these two
methods of influence. This is what the present
study was designed to do.
The differential effectiveness of preaching
459
460
J. PHILIPPE RUSHTON
vis-a-vis modeling in the studies above may
have been due to the ineffectiveness of the
manipulation of the preaching variable. Three
ways in which the preaching manipulation in
the previous studies appeared weak were in
relation to (a) their timing; (b) their direction; and (c) the power of the preacher.
As regards timing, while the modeling
always occurred when the model was in a
"win" situation, that is, collecting tokens for
himself and being presented with an actual
opportunity to donate, the preachings occurred either on "non-win" trials as in the
Bryan and Walbek (1970a, 1970b) studies or
before any trials at all as in the Grusec
(1972) and Grusec and Skubiski (1970)
studies. It is interesting that in the Bryan
(1971) study in which an interaction involving the preaching variable did reach significance, the timing of the preaching in that
study occurred immediately after obtaining a
winning score. The author, however, did not
refer to the timing when he sought to explain
his results and was unable to interpret the
interaction. It is suggested that the timing of
the preaching relative to that of modeling
may be crucial. The current study was
designed to test this.
As regards direction, in the Bryan and
Walbek (1970a, 1970b) studies, the exhortations were apparently delivered more to an
empty room than addressed to the child; and
in the Grusec studies the manipulation required the model to take "care not to look
at the child" (Grusec, 1972, p. 142). In the
current study, the preaching was directed at
the child so resembling more the natural
situation in which one seeks to exert
influence.
As regards power, Grusec (1971) has
shown that a powerful model produces more
imitation than a nonpowerful one, but none
of the studies varied the powerfulness of the
model when examining preaching. The present
study used a more powerful model than was
used in the previously mentioned studies.
Hence, the current study was designed to
test the differential effectiveness of modeling
and preaching, using the same experimental
procedure as Bryan and Walbek (1970a,
1970b) with the preaching manipulation
strengthened in the three ways outlined:
timing, direction of the preaching, and power
of the model.
A further issue of some importance is the
question of the durability and generality of
behavioral learning following exposure to a
model's actions and verbalizations. Apart
from a S-day retest of exposure to different
types of modeling situations (White, 1972),
there is virtually no evidence that exposure
to a model produces anything more than conformity to the demands of the immediate experimental situation (Krebs, 1970). Hence,
the current study employed a 2-month retest
of the effectiveness of the independent variables and also tested for generality across
dissimilar situations.
A still further issue to be investigated is
the part played in altruistic behavior by cognitive developmental variables (Kohlberg,
1969). Although social learning theorists acknowledge that observer attributes may
partly determine the efficacy of modeling influences (Bandura, 1969), rarely are such
attributes varied independently. This is particularly noticeable in the field of moral
development where such cognitive processes
as role taking and moral judgment have been
postulated to account for the age difference
in sharing behavior frequently reported
(Bryan & London, 1970; Rosenhan, 1969).
In a previous study (Emler & Rushton,
1974), it was found that children's responses
to Piaget's stories concerned with distributive
justice were highly significantly related to
their generosity scores. If such cognitive variables are indeed determinants of the behavior rather than just covariates, they might
be expected to interact with and affect the
reception of the social learning variables. A
third focus of the present study then was the
effect of children's level of moral judgment
on generosity under a variety of experimental
conditions.
METHOD
Subjects
Subjects were 140 children aged 7 to 11 years
from a predominantly working class area of London.
There were 28 children of each age, divided equally
between the sexes.
GENEROSITY IN CHILDREN
Design
A 2 X 2 X 3 factorial design was employed for 120
of the subjects with the remaining 20 being assigned
to a control condition. The factors were the child's
high versus low moral judgment; a model's generous
versus selfish behavior; and a model's generous
versus selfish versus neutral preaching. Within each
cell, subjects were equated for age and sex to permit
examination of these effects on generosity, too. Each
subject participated in two experimental sessions
separated by 8 weeks. Within each class, a subject
was sent to an experimenter in an order determined
by the teacher and then randomly assigned to treatment condition. On retest, all subjects within conditions, again equated for sex and age, were divided
into two equal groups and assigned to one of two
conditions, one in which the retest condition was
similar to the original testing situation and one in
which it was dissimilar.
Procedure
Initial testing. Each subject was sent by teacher
to the experimental room where an experimenter
met him and established a rapport. Six stories1
adapted from those used by Piaget (1932) were told
to the subject with a standard series of questions
to elicit the subject's concepts of moral justice. The
order was randomized across subjects. Subjects'
responses were recorded on tape for subsequent
analysis on the basis of Piaget's (1932) criteria.
The experimenter then showed the subject a
number of attractively arranged prizes ranging from
comics and small puzzles to games suitable for both
boys and girls of the age range. The subject was
asked if he would like to win one of these prizes.
When the subject said yes, the experimenter showed
him the electronic bowling game. This was similar
to that described by Rosenhan and White (1967),
being about 3 feet (.91m) long with an upright
panel at the end of the runway to indicate scores
on each trial. Four scores were possible—5, 10, IS,
and 20—the last being counted as a winning score.
The predetermined sequence included eight scores of
20 in a game of 20 trials. A buzzer sounded at the
termination of the game.
There were 100 tokens, in piles of 2, placed on
a table beside the game. Also on this table was a
bowl beneath a "Save the Children Fund" poster
depicting a poorly clothed child. A caption read
"Please Give." The bowl contained 10 tokens.
Another bowl was provided to hold subject's winnings. The experimenter introduced the subject to
1
Full details of the stories and coding procedures
can be found in the doctoral dissertation of the
author submitted to the University of London,
October 1973 entitled Social learning and cognitive
development: Alternative approaches to an understanding of generosity in 7 to 11 year olds. Alternatively, a mimeographed sheet of the stories will be
sent upon request to the author.
461
the same-sex model who was presented as a possible
future teacher in the school (this was to make the
model appear powerful) and was interested in playing
the game, too.
The experimenter explained to the subject and the
model how to play the bowling game, how to reward
themselves with tokens, and if they wished to, how
to donate to "Bobby," the child in the poster. The
optional nature of giving was stressed. They were
told that the more tokens they won, the better prize
they could have. The experimenter suggested that the
model should play the game first and instructed
the subject to watch. The experimenter then left
the room. The model then played the game and
either donated to Bobby one of each of the two
tokens he won on winning trials (generous model)
or he kept both for himself (selfish model). At the
same time he preached either generosity, selfishness,
or provided a neutral conversation. The preaching
was given on winning trials with the model looking
at the subject and saying with emphasis:
Generous preaching. "We should share our tokens
with Bobby" (winning Trials 1 and 5); "It's good
to give to kids like him" (winning Trials 2 and 6);
"It's right to share counters with Bobby" (winning
Trials 3 and 7); "You should give to kids like him"
(winning Trials 4 and 8).
Selfish preaching. "We should not share our
counters with Bobby" (winning Trials 1 and 5 ) ;
"It's not good to give to kids like him" (winning
Trials 2 and 6); "It's not right to share tokens with
Bobby" (winning Trials 3 and 7); "You should not
give to kids like him" (winning Trials 4 and 8).
Neutral preaching. "This is a nice game" (winning
Trials 1 and S); "I really like playing this game"
(winning Trials 2 and 6); "This is really fun"
(winning Trials 3 and 7); "I like this game"
(winning Trials 4 and 8).
When the buzzer sounded at the end of the game,
the model reset it and told the subject to play the
game now as it was his turn. The model then left
the room, ostensibly to take his winnings out to the
experimenter. When the buzzer sounded at the end
of the subject's game, the experimenter reentered
the room, counted up the tokens in the subject's
bowl, and told the subject prizes would be given out
after everyone had a turn at the game. The experimenter then asked the subject to indicate his view
of the model as very nice, just okay, not so nice,
or very bad and, further, how much he would like
to be in the model's class if the model became
teacher in that school, a lot, a little bit, don't mind,
or not so much.
Two-month retest similar condition. The same
male experimenter came to the class, called out the
subject's name, took him back to the same experimental room as before, asked him if he liked the
prize he had been given, and told him that since
there were some prizes left he could have another
go at the same game. The experimenter reminded
the subject of how to play the game and how to
win tokens saying only in passing that if the subject
J. PHILIPPE RUSHTON
462
wished to give any of his tokens to charity he could
still do so.
Two-month retest dissimilar condition, A female
experimenter the children had never seen before
came to the class and called out the subject's name
and took him to a different room in a different
part of the school and reminded him how to play
the game etc, In addition, the poster of "Bobby"
was changed from a yellow background depicting
one poorly dressed little English boy to a red one
depicting three obviously starving Asian children.
Controls. There were 20 control subjects, two
from each age of 7-11 divided evenly between the
sexes and randomly assigned to a control condition.
Subjects were administered the moral judgment
stories, as in the experimental group, but were not
exposed to the modeling situation. They were shown
how to play the game by themselves, were told how
to donate if they wished to, and then played immediately afterwards. For the retest conditions they
were divided into two equal groups and divided into
either the similar or dissimilar retest condition.
Experimenter bias controls. The experimenter
was kept blind as to the subject's test conditions.
In the immediate test this meant that it was the
model who selected the treatment condition and the
experimenter was blind as to which condition the
subject was in while he administered the moral
judgment stories and gave instructions.
In the retest condition, the experimenter simply
had the subjects' names on a list with no knowledge
as to which condition each subject had been run
under 2 months previously or how many tokens he or
she had donated. This was even more true for the
experimenter in the dissimilar retest condition who
had never met the children.
= 6.S4, p< .05; X 2 (4) = 14.67, p < .01.
Scores for the two stories were combined
and subjects divided into those scoring above
and those below the median. The age relationship to this combined moral judgment
score was significant, x 2 (4) = 14.92, p < .01.
There were no sex differences on these moral
judgment stories, either separately or combined, x 2 ( l ) = .30, X 2 (l) = -30, X 2 (D - -65.
A preliminary analysis of variance was carried out to test the effects of age and sex on
immediate generosity, retested generosity, and
evaluations of model. The measure of generosity consisted of the number of tokens the
subject donated to the charity out of the 16
won. No significant age differences were found
on the immediate generosity or upon the subject's evaluations of the model. There was,
however, a highly significant age difference on
donating behavior on retest, F(4,106) = 7.8,
p < .001, such that older children gave more
than younger children (see Table 1).
As there were no significant sex differences
nor Sex X Age interactions on immediate
generosity, retested generosity, or evaluations
of the model, all further analyses were collapsed over sex.
RESULTS
The effects of modeling, preaching, and
moral judgment were then examined against
the immediate generosity scores. Modeling
was found to be a highly significant determiner of the subject's immediate donations,
F(\, 104) = 158.18, p < .001. Preaching had
no effect, F(2, 104) < 1.00. The subject's
level of moral judgment was also significant,
F( 1,104) = 7.47, p < .007. The mean number of tokens donated by subjects with low
and high moral judgment scores were 3.3 and
4.6, respectively. When age was introduced as
a covariate, however, moral judgment had
a somewhat weaker effect on donations,
F(l,103) =4.04, / > < .05. There was only
one, near significant, interaction, Preaching
X Modeling, F(2,104) = 2.74, p < .07. Generous models who preached selfishness produced less donating than did generous models
who had preached generosity, or who had
made neutral comments. Further, selfish
models who preached neutral comments pro-
Four experimental subjects and two control
subjects were dropped from the analyses due
to their failure to turn up for the retesting.
Of the six moral judgment stories given to
subjects, only two showed the positive relationship with age necessary to be considered
as a cognitive-developmental variable, \ 2 (4)
TABLE 1
MEAN NUMBER OF TOKENS DONATED BY AGE,
SEX, AND TIME OF TESTING
Age (in years)
Immediate test
Boy
Girl
Two-month retest
Boy
Girl
7
g
9
10
11
3.8
2.0
3.9
4.5
4.2
4.8
4.0
3.9
5.4
5.1
2.6
2.9
2.8
1.7
3.6
4.7
4.4
2.9
7.3
5.8
Generosity: Immediate Test
463
GENEROSITY IN CHILDKEN
TABLE 2
MEAN NUMBER or TOKENS DONATED IN IMMEDIATE
TEST BY MODELING AND PREACHING
Preaching
Generous
Selfish
Generous
Neutral
Selfish
7.6
1.8
7.6
.9
6.1
1.7
Note. Controls donated a mean number of 4.1 tokens.
duced less giving than selfish models who
preached either generosity or selfishness.
These results are shown in Table 2.
In order to test the possibility that modeling or preaching might have different effects
depending on the subject's age, a further analysis of variance was computed. Neither modeling nor preaching, alone or in combination,
produced significant interactions with age,
F = 1.32, F = 1.18, F < 1.00, respectively.
Evaluations of Model
The two different measures of subject's
evaluations of the model (liking rated 1-4;
and desire to be in the model's class if the
model became a teacher in that school rated
1-4) were virtually identical. Results are
therefore shown only for the second measure,
unless otherwise stated. Both the model's
preaching and behavior were highly significant determiners of his judged attractiveness, F(2,104) = S.62, p < .005; F(l, 104)
= 21.19, p < .001, respectively. The Preaching X Behavior interaction was also significant such that the selfish model who preached
neutral comments but did not donate was
judged more attractive than the selfish model
who had preached either generosity or selfishness, F(2,104) = 4.49, p < .01. Table 3 presents the ratings of the model. Scores are
such that the higher the score, the greater
the model's attractiveness.
Moral judgment had no main effect on the
subject's evaluation of the model, F( 1,104)
< 1.0, but did interact with the preaching
manipulation on the second measurement of
the model's attractiveness, F(2,104) = 3.32,
p < .05, although not on the first measure
(F < 1.00). This relationship held when age
was covaried, F(2,103) = 3.41, p< .OS. On
this second measure, subjects with a high
moral judgment score rejected the selfish
preacher (.£ rating = 2.3), while those with a
low moral judgment score did not (X rating
= 3.0).
Generosity: Two-Month Retest
A comparison of the dissimilar and similar
retesting situations was made by carrying out
an analysis of variance using the retesting
condition as an independent variable in conjunction with the modeling and preaching
conditions. The type of retest condition
yielded no significance, either as a main effect, F (1,103) < 1.00, or in interaction with
modeling, F (1,103) < 1.00, preaching, F (2,
103) < 1.00, or both F (2,103) = l.SS. Thus
further analyses were collapsed over retest
conditions. Thus, the long-term effects of
modeling, preaching, and moral judgment
were able to be examined against the combined 2-month retest scores.
Even after an 8-week gap, modeling was
still a highly significant, F (1,104) = 16.44,
p < .001, influence on subject's donating behavior. In addition, preaching which had no
effect in the immediate test now had a highly
significant overall effect, F (2,104) = 4.77,
p < .01. Models who preached selfishness produced less giving than those who preached
generosity or neutral messages. Further, there
was a significant Preaching X Modeling interaction effect, F (2,104) = 3.48, p < .OS, such
that the model preaching selfishness but behaving generously caused subjects to donate
the least number of tokens in the generous
model condition, while the model preaching
generosity but behaving selfishly produced the
most giving in the selfish model condition.
Furthermore, the model preaching neutral
messages in the generous model condition
produced more giving than the generous model
TABLE 3
MEAN RATINGS OF MODEL'S ATTRACTIVENESS DY
MODELING AND PREACHING
Preaching
Model
Generous
Selfish
Generous
Neutral
Selfish
3.7
2.5
3.3
3.2
3.2
1.8
J. PHILIPPE RUSHTON
464
TABLE 4
MEAN NUMBER OP TOKENS DONATED IN RETEST
BY MODELING AND PREACHING
Preaching
Generous
Selfish
Generous
Neutral
Selfish
5.2
3.2
6.8
2.8
3.3
2.4
Note. Controls denoted a mean number of 4.2 tokens.
preaching generosity. These results are shown
in Table 4.
Differences in donating behavior for subjects with high and low moral judgment
found in the first test did occur on retest,
but only when age was not covaried from
the analysis, F (1,104) = 8.SS, p < .004.
With age introduced as a covariate, the significant relationship between moral judgment and
generosity disappeared, ^(1,103) = 2.42.
Moral judgment did not exert any significant
interaction effect.
Control subjects donated a mean number of
4.2 tokens on the retest, there being a correlation of r = .71 (df = 16, p < .001) between
their behavior on Test 1 and that on Test 2.
The consistency of behavior between the first
test and retest 8 weeks later was also considerable for the experimental group, r (114) =
.51, p< .001.
DISCUSSION
Modeling has been shown, in the context of
this experiment at least, to be a powerful
determiner of children's sharing behavior.
Further, this behavior evidences both durability and generality. Thus, according to
Kreb's (1970) criteria, it is reasonable to
conclude that new learning or internalization
of this moral behavior has occurred. It would
seem that a model does more than simply
induce temporary situational conformity mediated by some vaguely defined form of "demand characteristics" or "experimenter effect." If the usual demand characteristics
explanation were viable, one would expect the
model's preachings to work as an additional
cue to the child, especially when preaching
and modeling contradicted one another. This,
in the immediate test, was not the case, despite our attempts to strengthen substantially
the preaching manipulation. This finding thus
replicates the results reported by Bryan and
Walbek (1970a, 1970b).
The effectiveness of the preaching manipulation can hardly be doubted given that (a)
preaching was a powerful source of the child's
evaluations of the model and (b) preaching
did have a major effect on the 2-month follow-up study.
The finding that while preaching did not
directly influence subject's behavior it did
influence subject's evaluations of the model
again replicates Bryan and Walbek (1970a).
In addition, however, this study demonstrated
a significant Preaching X Modeling interaction on evaluations (see Table 3) which
Bryan and Walbek (1970a, 1970b) failed to
find. Selfish models who said "This is really
fun" were seen as more attractive than selfish
models who preached generosity or selfishness.
This suggests that rather than the model providing a "neutral" preaching, he provided
a source of positive affect. Table 2 provides
some support for this suggestion. Imitation of
the model was highest, for both generous and
selfish models, in the neutral condition. Table
4 provides further support. The significant
Modeling X Preaching interaction on retest is
partially due to the generous model preaching
neutral messages producing the most donating
in the generous model condition. Model affect
consisting of the model saying "I am happy"
has previously been shown to increase subsequent imitative generosity (Bryan, 1971),
particularly when immediately following the
model's behavior. Bryan (1971), however,
failed to find that imitative measures were related to the subject's judgments of the model.
Bryan also, however (Bryan & Walbek, 1970a,
1970b), failed to find the subject's behavior
related to the model's preaching. The discrepancy between the findings reported here and
those reported by Bryan is probably due to
the increased strength of the preaching manipulation in the present study, which
strengthened preaching in respect to (a) its
timing; (b) its direction; and (c) the power
of its deliverer.
The strength of the preaching manipulation emerged strongest in regard to the 2month retest. This "delay effect" is highly
interesting and appears to be due to the
GENEROSITY IN CHILDREN
preachings differentially affecting the decay
rates of the behaviors originally induced by
the modeling. Generally, when preaching was
consistent with the modeling, less decay or
regression to the mean occurred. On the other
hand, when preaching was discrepant with
the modeling, most regression to the mean
occurred. The explanation for this finding is
probably best sought at one or more of the
component subprocesses in the information
processing system governing modeling phenomena. These subprocesses include attentional mechanisms governing initial registration of modeled events; storage of the modeled events; retrieval of the events from
storage after 2 months; and the motivational
processes that determine whether the remembered events are activated into overt performance (Bandura & Jeffrey, 1973).
One possible explanation within this framework would be that in the initial test the perceptual dominance of the model's actions
caused the subject to rely predominately upon
his visual memory when playing the game
himself. This is what would be expected from
Bruner et al.'s (1966) work on the development of cognitive processes in children. Children of the age range used in this experiment
would be expected to be more attentive to
perceptual attributes of situations rather than
to symbolic attributes. However, both on the
2-month retest and on the experimenter's
questioning the subject of his evaluation of
the model, the subject is of necessity required
to rely more upon symbolic process and semantic memory either to recall the initial
situation to himself or to process the meaning
of the questions. Thus the model's verbalizations are perforce made more equally salient
to the model's actions.
An alternative explanation in terms of the
information processing system would be that
inconsistencies in stimulus input would cause
differential memory loss over a long period of
time although not over a very short period
due to the fact that memory loss in the longterm store appears to be primarily due to
semantic confusion (Baddeley & Dale, 1966).
Thus the child has greater difficulty retrieving
the relevant information in the discrepant
condition due to the competing memory
traces that were stored canceling each other.
465
In this study, independent measures of the
state of the information being processed at
different points were not taken. Ideally, future
research will attempt to elucidate the elements in the information processing system
responsible for the observed regularities in the
overt behavior.
That the delay effect was no chance occurrence is demonstrated by (a) the power of its
statistical association, both as a main effect,
and in interaction; (b) the rational patterning of that interaction; and (c) the finding
that exactly the same patterns of results occurred across retest situations. Words and
deeds appeared to have independent effects on
the children's behavior.
It would appear that while the social learning paradigm is relatively explicit and the
conditions under which models exert maximal
effects are fairly well understood, the relation
of verbal communications to behavior requires much more extensive investigation.
On the basis of earlier work (Emler &
Rushton, 1974; Rubin & Schneider, 1973) it
was expected that predictions of a child's
generosity could be made with better than
chance results from a knowledge of the child's
level of moral judgment. This expectation
was confirmed in the present study where an
association was found between moral judgment and generosity both on the immediate
test and on the delayed test. However when
age was covaried from the analysis the
strength of the relationship was weakened.
Indeed on the 8-week retest, covarying age
reduced the relationship to below an acceptable level of statistical significance. In some
ways this is surprising. That moral judgment
might show only a weak effect in the immediate test when the manipulation of modeling was having its maximum effect is understandable if it can be argued that the
modeling effect was blanketing all other effects (an idea that is supported by the lack
of normally expected age relationships with
the immediate test of generosity). It would
then have been expected that in the retest
condition, when any enduring individual characteristics would reemerge (as with the age
relationship to generosity), moral judgment
would exert a stronger effect. However this
was not found.
466
J. PHILIPPE RUSHTON
On the question of causality, the results
might suggest that moral judgment did not
have causal influences on the generosity scores
of the children. This argument would rest on
the assumption that if moral judgment had
been a determiner of generosity, rather than
just a covariate, it would be expected to interact with, and affect the reception of, other
determiners of altruistic behavior. The behavior of a model, whether behaving generously
or selfishly, might be expected to have had a
differential effect depending on whether it
was observed by a child with a high or a low
level of moral judgment. Such interactions did
not occur. It should be noted, however, that
there was a significant Preaching X Moral
Judgment interaction on the child's evaluation
of the model's attractiveness. Children with a
high moral judgment score rejected the selfish
preacher while those with a low moral judgment score did not. Thus moral judgment was
exerting an effect. Finally, this study demonstrates the value of examining social learning
and cognitive developmental variables over
time and across situations.
REFERENCES
Baddeley, A. D., & Dale, H. C. A. The effect of
semantic similarity on retroactive interference in
long and short term memory. Journal of Verbal
Learning and Verbal Behavior, 1966, 5, 417-420.
Bandura, A. Principles of behavior modification.
New York: Holt, Rinehart & Winston, 1969.
Bandura, A., & Jeffrey, R. W. Role of symbolic coding and rehearsal processes in observational learning. Journal of Personality and Social Psychology,
1973, 26, 122-130.
Bruner, J. S., et al. Studies in cognitive growth.
New York: Wiley, 1966.
Bryan, J. H. Model affect and children's imitative
altruism. Child Development, 1971, 42, 20612065.
Bryan, J. H., & London, P. Altruistic behavior by
children. Psychological Bulletin, 1970, 73, 200-211.
Bryan, J. H,, & Walbek, N. H. Preaching and practicing generosity: Children's actions and reactions.
Child Development, 1970, 41, 329-353. (a)
Bryan, J. H., & Walbek, N. H. The impact of words
and deeds concerning altruism upon children.
Child Development, 1970, 41, 747-757. (b)
Emler, N. P., & Rushton, J. P. Cognitive-developmental factors in children's generosity. British
Journal of Social and Clinical Psychology, 1974,
13, 277-281.
Grusec, J. E. Power and the internalization of selfdenial. Child Development, 1971, 42, 93-105.
Grusec, J. E. Demand characteristics of the modeling
experiment: Altruism as a function of age and
aggression. Journal of Personality and Social Psychology, 1972, 22, 139-148.
Grusec, J. E., & Skubiski, S. L. Model nurturance,
demand characteristics of the modeling experiment and altruism. Journal of Personality and
Social Psychology, 1970, 14, 3S2-359.
Kohlberg, L. Stage & sequence: The cognitive-developmental approach to socialization. In D. A.
Goslin (Ed.), Handbook of socialization theory
and research. Chicago: Rand McNally, 1969.
Krebs, D. L. Altruism—An examination of the concept and a review of the literature. Psychological
Bulletin, 1970, 73, 258-302,
Piaget, J. The moral judgment of the child. London: Routledge, 1932.
Rosenhan, D. Some origins of concern for others.
In P. Mussen, J. Langer, & M. Covington (Eds.),
Trends and issues in developmental psychology.
New York: Holt, Rinehart & Winston, 1969.
Rosenhan, D., & White, G. M. Observation and rehearsal as determinants of pro-social behavior.
Journal of Personality and Social Psychology,
1967, J, 424-431.
Rubin, K. H., & Schneider, F. W. The relationship
between moral judgment, egocentrism, and altruistic behavior. Child Development, 1973, 44, 661665.
White, G. M. Immediate and deferred effects of
model observation guided and unguided rehearsal
on donating and stealing. Journal of Personality
and Social Psychology, 1972, 21, 139-148.
(Received July 30, 1973)